Parked vehicle detection and status evaluation on X-band spotlight-mode SAR interferometry

For urban area monitoring, detection and classification of small targets with synthetic aperture radar imagery are one of the useful solutions. The conventional method has the problem that when a target on both master and slave images at the same position has the same radar cross section, it does not classify the target as either a slightly moved reparked vehicle or as a different vehicle. To overcome this problem, we propose a Fourier-transform-based simple phase compensation method for both accurate coherence estimation and classification of the reparked vehicle's status. The proposed method obtains the maximum value of the frequency basis, and also determines the threshold, which is based on effective sample number, with a constant false alarm rate. This number is determined by the peak signal-to-noise ratio in estimation window, and are used on the preprocessed probability density function for each sample number. Finally, the proposed method classifies a target as either a slightly moved vehicle or as a different vehicle by using both the proposed coherence and threshold detection, which enables it to determine the target status. Experimental results with the COSMO-SkyMed X-band spotlight SAR images demonstrate the effectiveness of the proposed method.